Download updated version

Focus-related constituent order variation without the NSR:
A prosody-based crosslinguistic analysis*
Daniel Büring & Rodrigo Gutiérrez-Bravo
Abstract
In this paper we show how to derive focus-related constituent order variation
in transitive clauses in English Spanish, and German, within a prosodic theory of
prominence such as Selkirk (1995) or Truckenbrodt (1999). These works have
noted that focus, intonation and constituent order requirements can be in conflict
with one another, and Truckenbrodt (1999) has shown that sacrificing the
canonical pattern of phonological phrasing is a strategy that some languages use
to solve such potential conflicts. Building on these observations, we propose that
movement of focused XPs is the strategy that is used to accommodate these
conflicts in languages where sacrificing the canonical pattern of phonological
phrasing is not an option. The resulting analysis is one where the focus and
intonational requirements of English, Spanish and German are essentially the
same, contrary to what has been suggested in previous work. Instead, these
languages differ in when and whether they sacrifice phonological phrasing
requirements or syntactic requirements. As such, our analysis preserves the
insights captured in Zubizarreta's (1998) treatment of these phenomena, while
maintaining the assumption, shared in most recent work on phonological
phrasing and accenting, that prominence is not governed by parameterized
syntax-based principles such as the Nuclear Stress Rule.
0. Introduction
It is well known that languages differ in whether or not they show noncanonical constituent order in cases of subject focus (Contreras 1976, Vallduví
1992, Ladd 1996, Zubizarreta 1998, to cite just a few references), a phenomenon
that will be referred to henceforth as focus-related constituent order variation.
There are languages like English (SVO), where constructions with focused
subjects display canonical constituent order, with a focused subject receiving the
main pitch accent (marked by caps) in-situ, as in (1).
(1) a. Who bought the newspaper?
b. JOHN bought the newspaper.
SVO
On the other hand, there are languages like Spanish (also SVO), that display
subject inversion when the subject is in focus, as in the answer in (2b).
2
(2)
Spanish
a. Quién compró el periódico ayer?
who bought the newspaper yesterday
‘Who bought the newspaper yesterday.’
b. Ayer
compró el periódico JUAN.
yesterday bought the newspaper Juan
‘JUAN bought the newspaper yesterday.’
VOS
Finally, there are languages like German that display both patterns. Thus in
German subordinate clauses (canonically SOV) a focused subject can be
accented in-situ, as in the answer in (3b), or it can follow the object, as in (3c)
(although there is a slight preference for the former construction).1
(3)
German
a. Wer ernnent
den Aussenminister? - Es wird erwartet, dass
who nominates the foreign-minister it is
expected that
‘Who’s in charge of nominating the foreign minister?- It is expected
that…’
b. der KANzler den Aussenminister ernennt.
SOV
the chancellor the foreign-minister nominates
‘…the chancellor nominates the foreign minister.’
c. den Aussenminister der KANzler
ernennt.
OSV
the foreign-minister the chancellor nominates
‘…the chancellor nominates the foreign minister.’
On the other hand, in transitive sentences in which the direct object, the VP,
or the whole sentence is in focus, all three languages behave similarly, with the
main pitch accent on the direct object of the clause, and the subject preceding the
object as shown in (4-6).
(4)
English
a. Sentence focus
i. What’s been happening?
ii. John bought the NEWSpaper.
b. VP focus
i. What did John do?
ii. John bought the NEWSpaper.
c. Object Focus
i. What did John buy?
ii. John bought the NEWSpaper.
3
(5)
Spanish
a. Sentence focus
i. Qué pasó?
‘What happened.’
ii. Juan compró ayer
el peRIÓdico.
Juan bought yesterday the newspaper
‘Juan bought the newspaper yesterday.’
b. VP focus
i. Qué hizo Juan?
‘What did Juan do?’
ii. Juan compró ayer
el peRIÓdico.
Juan bought yesterday the newspaper
‘Juan bought the newspaper yesterday.’
c. Object Focus
i. Qué compró Juan ayer?
‘What did Juan buy yesterday?’
ii. Juan compró ayer
el peRIÓdico.
Juan bought yesterday the newspaper
‘Juan bought the newspaper yesterday.’
(6)
German
a. Sentence focus
i. Worauf
warten all? - Es wird erwartet, dass
where-upon wait
all it is
expected that
‘What’s everybody waiting for? -It is expected that…’
ii. der Kanzler
den AUssenminister ernennt.
the chancellor the foreign-minister nominates
‘… the chancellor nominates the foreign minister.’
b. VP focus
i. Was wird der Kanzler
als nächstes tun? -Es wird erwartet,
what will the chancellor for next
do it is
expected
dass
that
‘What’s the chancellor supposed to do next?-It is expected
that…’
ii. der Kanzler
den AUssenminister ernennt.
the chancellor the foreign-minister nominates
‘… the chancellor nominates the foreign minister.’
c. Object Focus
i. Wen ernennt
der Kanzler
als nächstes? - Es wird
who nominates the chancellor for next
it is
erwartet, dass
expected that
‘Who is the chancellor going to nominate next? -It is expected
that…’
ii.
der Kanzler
den AUssenminister ernennt.
the chancellor the foreign-minister nominates
‘… the chancellor nominates the foreign minister.’
4
Descriptively, what the examples in (4-6) show when compared with the
examples in (1-3), is that all three languages have fairly similar intonational and
syntactic properties. When the subject is in focus, however, these languages differ
in whether the intonational pattern is changed, as in English, whether canonical
constituent order is sacrificed, as in Spanish, or whether either possibility is
allowed, as in German.
In a large number of recent works (Vallduví 1992, Ladd 1996, Zubizarreta
1998, to cite just a few references; see also Costa 1996, 1998 for an Optimalitytheoretic approach), focus-related constituent order variation as in (2b) and (3c)
has been linked to the need for foci to receive prominence via sentential stress,
sentential stress in turn determining the location of the nuclear pitch accent; in
other words, the target position of the focused items is not a specific focus
position, but simply a (nuclear) stress position. The difference between focus
marking between, say, English and Spanish is not ‘prosodic prominence versus
syntactic position’, but rather ‘prosodic prominence in situ versus prosodic
prominence in a specific position’. Focus marking, on this view, is inevitably
realized by prosodic prominence, in particular stress.
The present proposal
shares this view, but differs from
previous
implementations of it (such as Zubizarreta 1999, Neeleman & Reinhart 1998,
Costa 1996, Choi 1996) in that it adopts a genuinely prosodic account of stress.
It thereby attempts to bridge the gap between the aforementioned theories of
constituent order variation, which assume a syntax-based algorithm for main
stress assignment, and the huge phonological literature on phrasing and stress,
which convincingly shows that stress assignment is based on prosodic structure.
This in turn yields empirical advantages, such as a uniform treatment of nuclear
and pre-nuclear pitch accents, which ultimately results in a simpler theory of the
focus--stress relation.
5
Specifically, we propose to views stress assignment as a function of the
alignment of prosodic phrases, in particular at a level we call the phonological
phrase, which roughly corresponds to the accent domains of Gussenhoven
(1984). Sentences (4)-(6) display the optimal way for all three languages to align
prosodic phrases to the syntactic structure. The subject focus patterns in (1b) and
(3b) on the other hand present adjustments of the phonological phrasing, which
serve to guarantee relative prominence patterns between the constituents that are
consistent with the focus structure assigned.
We thus
provide
a
specific
and
perhaps
somewhat
unexpected
implementation of the idea expressed by Vallduví (1992), Ladd (1996), among
others, that the assignment of sentential stress is strictly sentence-final in Spanish,
but flexible in English (and German). On our view, sentential stress assigned
identically in all constructions in all three languages, but “hits”
different
constituents, depending on the number and alignment of phonological phrases.
The uniformity of primary stress assignment in (4)-(6) is due to this one uniform
sentential stress assignment. The differences in the subject focus patterns result
from more subtle differences at the level of phonological phrase construction.
We present our approach in terms of ranked and violable constraints
(Optimality Theory), which lends itself naturally to the prosodic treatment
proposed. It also allows for an alternative approach to cross-linguistic variation,
which instead of switching particular mechanisms on and off (such as
Zubizarreta's (1998) ‘metrical invisibility’) or replacing one by the other (such
as ‘main stress by focus’ vs. ‘main stress by structure’) simply varies the relative
strength of identical conditions across languages. This kind of approach implies
that even a condition that is apparently irrelevant in a particular language, such as
the requirement to map onto a phonological phrase and hence give phrasal stress
to every lexical projection of a certain size --- which is clearly violated in subject
6
focus constructions in English --- can be seen to be active under certain
favorable circumstances in that language, for example broad focus constructions.
We believe that this pattern is indeed evidenced in the languages under
consideration quite in general, lending support to a constraint-ranking approach.
For example, the prosodic patterns of sentences with foci in medial position in
English and German (e.g. on the direct object in double-DP ditransitive clauses)
is identical to that in broad focus constructions to the left of the main stress,
though not to its right. This follows from the kind of approach given here,
though space does not permit us to discuss these more complex constructions
here.
In general, we have deliberately chosen to present an extremely limited
fragment of constructions in an extremely limited set of languages, in order to be
able to illustrate the essential workings of the prosody based account within the
confines of a journal article. More detailed investigations into the languages
discussed here can be found in Büring (2001a,b) and Gutiérrez-Bravo (2000).2
1. Constituent order
With respect to the syntax of the data under consideration, we will use the
following constraint, which establishes that the subject is structurally more
prominent than the object in the unmarked case.
(7) SO
The subject is structurally more prominent than the object.
-Violated when the subject does not asymmetrically m-command the
object.
Concretely, SO is satisfied when the subject occupies Spec-IP and the object
remains
in
its
VP-internal
position.
Crucially,
all
the
structures
7
in (8) violate this constraint. Whether the subject remains in Spec-VP (8a), or it is
right-adjoined to VP (8b), or the object is scrambled, the subject fails to
asymmetrically m-command the object.
(8)
a.
b.
VP
DP
Subject
VP
V’
V
VP
DP
Object
Spec
DP
Subject
V’
V
c.
DP
Object
VP
DP
Object
VP
DP
Subject
V
We follow the standard assumption that the verb undergoes movement to I
in Spanish.3 After V-to-I movement, (8b) and (8c) result in the VOS order
characteristic of the cases where the subject is in focus. We also assume that
German OSV is the result of scrambling of the direct object, where the subject
stays in Spec-VP, so these structures will also incur in a violation of SO.
While the effects of SO follow to a large extend from simpler and
independently proposed constraints such as Grimshaw's (1997)
SUBJECT
constraint or Costa's (1996, 1998) SUBJECT-CASE, some structural variants of
OS-ordering punished by SO, e.g. scrambling of the object across a subject in
Spec-IP in German, do not, and would hence necessitate the inclusion of further
syntactic constraints. Furthermore, a number of analyses of the syntax of Spanish
(Contreras 1991, Ordóñez & Treviño 1999, Gutiérrez-Bravo 2002) have claimed
that in this language the subject does not move to the preverbal position because
8
of Case considerations. Since these considerations are orthogonal to the main
goal of this paper, we will stick to the SO constraint for expository reasons.4
2. Phrasing and intonation
Throughout we assume that sentences are mapped into exhaustive strings of
phonological phrases (henceforth, pP’s), which in turn form an intonational
phrase (henceforth, iP). However, we assume that the languages
under
consideration differ with respect to the kinds of constituents which they allow
phonological phrases to correspond to. While English and German phonological
phrases sometimes contain more than one prosodic word and correspond
roughly to lexical XPs (Selkirk 1984, Truckenbrodt 1999), Spanish strictly
aligns phonological phrases with prosodic words (PrWds).5 So English and
German map a transitive sentence as in (9a) and (9b) respectively, whereas
Spanish maps it as in (10).
(9)
(10)
a. English
(S) (V
O )
[ NP] [
[ NP]VP]
b. German
(S)(
O V)
[ NP] [VP [NP ] ]
Spanish
( S ) ( V )( O )
[ N NP][ V [ N NP]VP]
The effect of these different mappings is subtle but systematic: Since each
phonological phrase has exactly one head, it has exactly one position with
phrasal stress that a pitch accent (nuclear or pre-nuclear) can be associated with.
Accordingly, we find two pitch accents in English and German transitive
sentences – one on the subject, one on the object – while there are three –
subject, verb, and object – in Spanish.
9
The well-known fact that predicate and argument form an ‘accent domain’,
or that focus can ‘project’ from an argument to a head, or that predicates and
argument ‘integrate’ (Schmerling 1976, Gussenhoven 1983, Rochemont 1986,
Jacobs 1993, Selkirk 1995, among many others) in English and German is thus
captured as a fact about prosodic phrasing.6 It is also worth noting that Spanish is
not the only language in which verbs do not integrate with their arguments and
hence bear their own pitch accent. A particularly striking example, brought to
our attention by a reviewer, is Bengali, as discussed in Hayes and Lahiri (1990),
where, too, the verb carries it own pitch accent. Since Bengali is a head-final
language, this pitch accent often becomes the nuclear pitch accent (i.e. the head
of the intonational phrase) as well.
We follow Truckenbrodt (1999) in assuming that (all else being equal)
phonological phrases do not align with functional projections, so we do not
represent in the schemas the functional projections present in the structures in (9)
and (10). From here onwards we represent phonological
phrases with
parentheses.
2.2 English and German
For the cases of simple transitive clauses, the phrasing pattern of English and
German in (9) can be derived immediately by Truckenbrodt’s (1999) constraints
WRAP and STRESS-XP: The former requires that lexical XPs not be ‘split up’ into
several prosodic phrases; the latter requires that each XP receive phrase level
stress, i.e., that it contain the head of a prosodic phrase, which effectively stops
prosodic phrases from growing too large. So if two lexical XPs exclude one
another, as in the case of the subject NP and VP, each forms its own prosodic
phrase, satisfying STRESS-XP; WRAP is not violated, since there is no higher
lexical XP containing the subject NP and the VP that gets split up by this
phrasing.
10
On the other hand, if a lexical XP contains another one (call it YP), WRAP
will demand that they be part of the same prosodic phrase. Crucially, STRESS-XP
can be satisfied in that configuration too if the head of that prosodic phrase (i.e.
the phrasal stress) is within the embedded YP (since in this case both XP and YP
receive phrase level stress; note that there is no constraint requiring that the
syntactic head of XP be itself stressed). Thus a VP containing an object will be
mapped onto one prosodic phrase, with the stress put on the object. Again both
STRESS-XP and WRAP are satisfied. The English/German phrasing and accenting
pattern is thus derived, regardless of the relative ranking of these two constraints.
WRAP and STRESS-XP, however, do not cover all cases which accent domain
formation rules such as Gussenhoven’s (1983) Sentence Accent Assignment
Rule (SAAR) do, for example. Observe that a number of predicate argument
constructions in English show the characteristic accent-on-the-argument pattern,
in which, however, the predicate is itself (or contains) a lexical XP, not just a
head:
(11) a. A TRAIN arrived. (unaccusative verbs)
b. The SUN is shining. (certain unergative intransitives)
c. They bought the SHOP empty. (resultative secondary predicates)
d. I’ve got a TRAIN to catch. (short infinitival relatives)
e. He met a GIRL he knew. (short finite relatives)
We conclude that the phrasing of verbs and their objects is not purely a
function of phrase structure, but rather of their semantic relation, and therefore
adopt the rather more descriptive constraints in (12) in favor of Truckenbrodt’s
constraints:
11
(12)
XP=pP:
Align a (lexical) XP with a phonological phrase.
A. PRED
A predicate shares a phonP with at least one of its arguments.
B. XP
A phonP contains an XP. If XP and YP are within the same phonP,
one contains the other (where X and Y are lexical categories).
The evaluation of the XP=pP constraint is exemplified in the tableaux (13)
and (14), where for clarity each condition is presented separately on the right
side of the tableau. Candidates (13b) and (14b), where there is a single
phonological phrase in the sentence, violate the XP condition, since the subject
and the object are part of the same phonological phrase but neither contains the
other. On the other hand, candidates (13c) and (14c), where every phonological
phrase corresponds to a prosodic word, violate the PRED condition, since the verb
does not form a phonological phrase with any of its arguments. The winning
candidates are thus (13a) and (14a), with one phonological phrase aligned with
the subject NP and the other one with the VP.
(13) English
XP=pP
PRED
XP
F a. (John) (bought the newspaper)
b. (John bought the newspaper)
c. (John) (bought) (the newspaper)
*!
*!
(14) German: ‘…(that) the chancellor nominates the foreign minister.’
XP=pP
PRED
XP
F a. (der Kanzler) (den Aussenminister ernennt)
b. (der Kanzler den Aussenminister ernennt)
c. (der Kanzler) (den Aussenminister) (ernennt)
*!
*!
12
When a phonological phrase contains both a predicate and its argument, as
in the VPs in (13) and (14), it is the argument that receives the accent, all else
being equal. This result is obtained by means of the constraint in (15):7
(15) ARGUMENT-OVER-PREDICATE (A/P)
Within a Phonological Phrase, an argument is more prominent
than a predicate.
Tableau (16) exemplifies for an English VP how this constraint is evaluated.
Concretely, A/P is violated when the verb is accented instead of its complement.
(16) English
A/P
F a. (bought the NEWSspaper)
b. (BOUGHT the newspaper)
*!
To be sure, the phrasing derived by these constraints only shows up if
focusing doesn’t intervene, as it does in the narrow subject focus cases we discuss
below. Other instances in which the phrasing will come out differently include
narrow verb focus and wide foci with given objects (e.g. pronouns), in all of
which cases the pitch accent falls on the verb. See Büring (2001a,b) for
discussion.
2.3 Spanish
Consider now Spanish, where, following an observation by Sosa (1991:54),
we propose that every phonological phrase corresponds to a prosodic word. This
kind of phrasing is the result of the constraint in (17).
13
(17)
PrWd=pP
Align the right edge of each prosodic word with the right edge
of a PhonP.
Violated once for every lexical head (N, V, etc.) whose right
edge is not aligned with the right edge of a PhonP.
Evidence for the claim that in Spanish every prosodic word corresponds to a
phonological phrase can be found in the fact that in Spanish every prosodic
word receives an accent (Fant 1984; Sosa 1991, 1999; Face 2000).
(18)
A+
A
A
B
Los mozos sacaron la moto
del garaje.8
the butlers took
the motorcycle of-the garage
‘The butlers took the motorcycle out of the garage.’
(Fant 1984)9
In (18) we adopt Fant’s representation of pitch accents in Spanish to
abstract away from dialectal variation; Sosa (1999), for example, analyses the A
accents in Latin American and Peninsular Spanish as L*+H, and the B nuclear
accent as typically L*. In a more detailed study of Peninsular Spanish, Face
(2000) also analyzes the pre-final pitch accents as L*+H, but instead analyzes the
final pitch accent as L+H*. The important point, however, is that the pattern in
(18) is observed in one way or another in the data analyzed in all these works.
It is important to note that the accents in (18) are not simply the realization
of lexical stress in Spanish, a fact that is systematically pointed out in the
literature on intonation in this language (see for instance Fant 1984, Sosa 1991,
1999; Hualde 2002 and references therein). This is because, as noted in these
works, there are cases in Spanish where prosodic words are not accented, but still
receive lexical stress, which shows that there is no necessary one-to-one relation
between accenting and lexical stress. The precise characterization of these cases is
still poorly understood, but they are sometimes described as instances where the
unaccented word has less semantic !or functional ”weight” than what it would
usually have. They typically correspond to “light” verbs such as be and go,
14
transitive infinitival verbs followed by their directs objects, and the verbs of some
lexicalized expressions. In example (19) we present an example of such a
lexicalized expression, from Sosa (1999: 131; gloss and translation are our own),
where we represent all lexically stressed syllables in small caps. Although Sosa
does not include this example as part of his discussion of the difference between
lexical stress and accent in Spanish, our observation is that (19) is not literally a
petition for someone to go and bring a chair closer, but rather a polite request
for someone to have a seat. In this case the unit formed by the verb and its
nominal complement receives just one accent, which falls on the complement.
(19)
L*+H L+H*
H*+H L%
|
|
|
QueRRÍA ustED acerCAR una SIlla?
would
you
bring.close a chair
‘Would you please have a seat?’
In analyzing the intonational pattern of Spanish, Alcoba & Murillo (1998)
propose that an intonational phrase in Spanish is composed of one or more tonic
groups, each of which consists of a prosodic word plus all the clitics and stressless
elements that appear to its left (Cf. also Sosa’s 1991, 1999 notion of rythmic
group). We would like to suggest that what has been described in the literature on
intonation in Spanish as tonic groups are phonological phrases that result from
the ranking where PrWd=pP outranks XP=pP. This is shown in tableau (20); for
clarity, violations of PrWd=pP are signaled in the tableau with the category of the
prosodic word that fails to have its right edge aligned with the right edge of a
phonological phrase.
15
(20) Spanish: ‘Juan bought the newspaper.’
PrWd=pP
a. (Juan) (compró el periódico).
V!
b. (Juan compró el periódico).
N!V
F c. (Juan) (compró) (el periódico).
XP=pP
*
Candidate (20a) violates PrWd=pP because the verb compró ‘bought’, does
not form a phonological phrase by itself. Candidate (20b), where a single
phonological phrase corresponds to the whole of the sentence incurs two
violations of this constraint because neither the subject nor the verb is rightaligned with a phonological phrase. The winning candidate is (20c), where every
prosodic word corresponds to a phonological phrase. The winning candidate
violates XP=pP, specifically, the PRED condition, since the predicate does not
form a phonological phrase with any of its arguments, but this violation has no
effects because of the ranking PrWd=pP >> XP=pP. English and German, on the
other hand, display the opposite ranking, as exemplified for English in tableau
(21). This derives the phrasing pattern previously discussed for these languages,
where the sentence is mapped into two phonological phrases, corresponding to
the subject NP and the VP. Here the losing candidate (21b) loses to (21a)
because of its violation of the PRED condition, since the predicate does not form a
phonological phrase with any of its arguments.
(21) English
XP=pP
F a. (John) (bought the newspaper).
b. (John) (bought) (the newspaper).
PrWd=pP
V
*!
16
2.4 Prominence in the intonational phrase
To conclude this section, we consider how the most prominent prosodic unit
in the intonational phrase is determined. Again following Selkirk (1984, 1995),
Truckenbrodt (1999), inter alia, we assume that one of the phonological phrases
that make up the intonational phrase is signaled by the nuclear pitch accent as
the head of the intonational phrase. We propose that in all three languages under
consideration this process is governed by the alignment constraint in (22), which
requires the rightmost phonological phrase to become the head of the
intonational phrase (see Truckenbrodt 1999).
(22)
iP-Hd-right:
Align the right edge of every iP with the right edge of the PhonP
that is the head of the iP.
-Violated when the phonP that receives the nuclear accent is not
right-aligned with the iP.
Henceforth, in the prosodic representation we mark the head of the iP by an
x within the iP. The constraint in (22) will be violated by a structure like the one
schematized in (23), where the phonological phrase that contains this accent is
not right-aligned with the intonational phrase:
(23)
(
(
X
)(
)iP
)PhonP
We further propose that iP-Hd-right is undominated in all three languages.
As a summary of the discussion on intonation in this section, the rankings of the
intonational constraints presented in this section for English, German and
Spanish are presented in (24).10
(24)
a. English:
b. German:
c. Spanish:
iP-Hd-right, A/P, XP=pP >> PrWd=pP
iP-Hd-right, A/P, XP=pP >> PrWd=pP
iP-Hd-right >> PrWd=pP >> XP=pP, A/P11
17
At this point, we can already derive one of the observations made in the
introduction. Although English and German resort to a phrasing strategy
different from the one observed in Spanish, the rankings in (24) yield the result
that, all else being equal, in a transitive sentence in the three languages the
nuclear accent will fall on the direct object. This is shown in the tableaux in (25).
(25) a. English
iPHd-R
A/P
XP=pP
PrWd=pP
F a. (John) (bought the NEWSpaper)
V
b. (John) (BOUGHT the newspaper)
c. (JOHN) (bought the newspaper)
*!
V
*!
V
d. (John) (bought) (the NEWSpaper)
*!
b. German
iPHdR
A/P
XP=
pP
F a. (der Kanzler)(den AUssenminister ernennt)
b. (der Kanzler)(den Aussenminister erNENNT)
c. (der KANzler)(den Aussenminister ernennt)
*!
PrWd
=pP
V
V
*!
V
d. (der Kanzler)(den Aussenminister)(erNENNT)
*!
c. Spanish
iPHd-R
PrWd=pP
F a. (Juan) (compró) (el peRIÓdico)
A/P
*
b. (Juan) (comPRÓ) (el periódico)
*!
c. (JUAN) (compró) (el periódico)
*!
d. (Juan) (compró el peRIÓdico)
XP=pP
V!
The fact that, when all else is equal, all three languages will show the same
intonational pattern, despite the differences in phrasing, will provide us with the
basic foundations to understand how and when they differ and how and when
they behave in a similar way once focus considerations are introduced.
18
3. Focus
Before introducing focus considerations to the picture laid out so far, we
present our assumptions regarding foci. In what follows, we will consider only
cases where the focus is the constituent in the answer to a simple
wh-question that corresponds to the wh-phrase in the question, as in (26).
(26) a. Who screamed?
b. JOHN screamed.
Delimiting focus in this way is of particular importance when we consider the
case of Spanish. Subject to some dialectal variation and also to the degree of
definiteness and individuation of the subject DP, Spanish presents constructions
like (27a), where the focused subject appears in Spec-IP (from Zubizarreta
1998). It has been noted in the literature (Silva-Corvalán 1983, Fant 1984, and
Zubizarreta 1998) that in these cases the subject has intonational properties
associated with it that are different from those of the sentence-final subject of
VOS sentences like (2b)12. More importantly, (27a) also differs from (2b) in its
interpretation: It can only have a corrective interpretation, such as an answer to a
question like (27b), where the answer is overtly implied by means of a tag or
understood to be implied by the hearer.13 Whether the focus in (26b) and (27a)
correspond to two different kinds of foci, or whether some other variable is
responsible for the differences in intonation and interpretation between these two
examples, is an issue that will not be dealt with here.
(27) a. MARÍA me regaló la botella de vino.
Maríato-me gave the bottle of wine
‘MARÍA gave me the bottle of wine.’
b. Quién te
regaló la botella de vino? (Juan, verdad)?
who to-you gave the bottle of wine Juan right
‘Who gave you the bottle of wine?(Juan, right?).’
19
Now, to derive the effect by which the focus ends up being intonationally
the most prominent constituent in the sentence we propose the formalization of
Truckenbrodt’s (1995) FOCUS PROMINENCE constraint in (28), which we further
assume to be undominated in all three languages under consideration.14
(28)
FOCUS PROMINENCE (FP)
Focus is most prominent.
-If a is a prosodic constituent at level n which contains a syntactic node
that is F-marked, a is the head of the prosodic category at level n+1 that
contains a.
The effects of this constraint are exemplified in (29). When the focus is a
prosodic word, this prosodic word must become the head of the phonological
phrase that contains it, as in (29a). In turn, a phonological phrase containing a
focus must become the head of the intonational phrase, as schematized in (29b).
As with the iP-head, an x at the phonological phrase level above a prosodic word
marks that prosodic word as the head of the phonological phrase.
(29) a. (
(
X
)PhonP
X )( XFOC )PrWd
b. (
X
)iP
(
X ) (
X
)PhonP
( X )( X ) ( X )(XFOC )PrWd
In a nutshell, what the FP constraint ensures is that the nuclear accent in an
intonational phrase will fall on the focus, thus making it the most prominent
constituent in the sentence. Now the crucial observation at this point is that when
the subject is the focus in a transitive construction, in all three languages a
conflict results
between
the
requirements
iP-Hd-Right, as schematized in (30):
of
FOCUS PROMINENCE and
20
(30)
a. English: (
(
c. Spanish: (
(
)(
SFoc
)(
SFoc V
V O
)(
O
)iP
)PhonP
b. German: (
(
SFoc
)(
O V
)iP
)PhonP
)iP
)PhonP
On the one hand, iP-Hd-Right requires the rightmost pP to be the head of
the intonational phrase; on the other, FOCUS PROMINENCE requires the
phonological phrase that corresponds to the subject (and which is not the
rightmost pP) to become the head of the intonational phrase so that the focus can
be most prominent. There are two ways to solve this conflict without violating
iP-Hd-Right or FOCUS PROMINENCE. One is to violate the constraint that governs
the optimal phrasing pattern in these languages (XP=pP in English and German,
PrWd=pP in Spanish) so that there is a single phonological phrase, which is right
aligned with the intonational phrase and which has the focus as its prosodic head
(see also Truckenbrodt 1999). This possibility is schematized in (31).
(31)
a. English: ( X
)iP
( X
)PhonP
SFoc V O
b. German: ( X
)iP
( X
)PhonP
SFoc O V
c. Spanish: ( X
)iP
( X
)PhonP
SFoc V O
An alternative possibility is to respect all the relevant intonational and
phrasing requirements, but sacrifice the syntactic requirements of the SO
constraint instead (descriptively, sacrificing canonical constituent order). This
can be achieved by either having the subject in a position where it can form the
rightmost phonological phrase with the verb, as in (32a) for English and (32b)
for German, or by having the subject in the rightmost position forming its own
phonological phrase, as in (32c) for Spanish.
21
(32)
a. English: (
(
c. Spanish: (
(
X)iP
X)PhonP
O V SFoc
)(
O
)(
b. German:
(
(
X
)iP
)(X
)PhonP
O SFoc V
X )iP
)(X )PhonP
V SFoc
English chooses the strategy in (31), Spanish chooses the strategy in (32),
while German allows for either. As will be shown in the following and final
section, the correct result can be arrived at simply through the ranking of the SO
constraint relative to the constraint that governs canonical phrasing in the
languages under consideration.
4. A cross-linguistic analysis
The final rankings for the languages under consideration are presented in
(33), where <<>> indicates a constraint tie. For simplicity we leave the A/P
constraint out of the rankings and in the tableau to follow we present only
candidates that satisfy this constraint.
(33)
a. English:
b. Spanish:
c. German:
FocP, iPHdR, SO >> XP=pP >> PrWd=pP
FocP, iPHdR >> PrWd=pP >> SO, XP=pP
FocP, iPHdR >> {SO <<>> XP=pP} >> PrWd=pP
Consider English first. As mentioned in the introduction, English does not
modify its canonical constituent order in cases of subject focus.
(34)
a. Who bought the newspaper?
b. JOHN bought the newspaper.
c. *Bought the newspaper JOHN.
SVO
VOS
In terms of the discussion developed so far, English can thus be understood
as a language that sacrifices optimal phrasing in favor of canonical constituent
22
order, which is the result of the ranking SO >> XP=pP. The analysis is presented
in the tableau in (35). We assume that focus is marked in the input, and that this
marking cannot be deleted, due to an undominated FAITH-FOCUS constraint not
included in the tableaux.
(35) English subject focus
FOCP
F a. (JOHNFoc bought the newspaper)
b. (JohnFoc) (bought the NEWSpaper)
c. (JOHNFoc) (bought the newspaper)
d. (bought the newspaper)(JOHNFoc)
iP-HdR
SO
XP=pP
*
*!
*!
*!
Consider the losing candidates in (35) one by one. Notice first of all that all
these losing candidates display the optimal phrasing pattern of this language, so
XP=pP is respected by all of them. They all lose, however, as the result of
violating a higher ranked constraint. Candidate (35b), where the nuclear accent
falls on the direct object of the verb, fatally violates FOCUS PROMINENCE, since the
phonological phrase that contains the focus is not the head of the intonational
phrase (and consequently, the focus is not the most prominent constituent).
Candidate (35c) satisfies FOCUS PROMINENCE by mapping the subject onto its own
phonological phrase and making this pP the head of the intonational phrase.
This candidate, however, is ruled out because of its violation of iP-Hd-R, since the
head of the intonational phrase is not the rightmost phonological phrase in iP.
Lastly, candidate (35d) satisfies both FOCUS PROMINENCE
and
iP-Hd-R by
making the focused subject the final constituent of the sentence, but in the
constraint ranking of English the resulting violation of SO proves fatal.15 The
winning candidate is candidate (35a), which makes a single phonological phrase
23
out of the whole sentence and makes the subject the head of this phonological
phrase by assigning the nuclear accent to it.
Note that both candidates (35a) and (35c) have canonical constituent order
and a nuclear accent on the subject (= the focus). The difference is that (35c) has
two phonological phrases with a “shifted” iP-head, while the winner (35a) has
one “super-sized” pP. Since we take the presence of a phrase level stress to be a
sufficient condition for pitch accent assignment, (35c) thus predicts that the
second pP, (specifically the noun newspaper) can bear a pitch accent, albeit a
secondary one.
However, this is not the case: English, like German,
characteristically allows for secondary pitch accent preceding the nuclear pitch
accent, but never following it. We take this to be evidence that these languages
achieve a shift of the main accent by means of “de-phrasing”
post-focal
material (i.e. extending the right boundary of the focus pP all the way to the end
of the iP), rather than by shifting the iP head to the left. This is why iP-HdR is
crucially ranked above XP=pP.
Spanish, on the other hand, displays the opposite pattern, in which canonical
constituent order is sacrificed in cases of subject focus. The correct result, shown
in tableau (37), is achieved by ranking PrWd=pP, the constraint that dictates
phonological phrasing in this language, over SO.16
(36)
a. Who bought the newspaper yesterday?
b. #JUAN compró ayer
el periódico.
Juan bought yesterday the newspaper
c. Ayer
compró el periódico JUAN.
yesterday bought the newspaper Juan
‘JUAN bought the newspaper yesterday.’
SVO
VOS
24
(37) Spanish subject focus
FOCP
a. (JUANFoc compró el periódico)
b. (JuanFoc) (compró) (el peRIÓdico)
c. (JUANFoc)(compró) (el periódico)
F d. (compró) (el periódico) (JUANFoc)
iP-HdR
PrWd=pP
SO
N!V
*!
*!
*
Candidates (37b) and (37c) are ruled out in the same way as candidates
(35b) and (35c) in English; the former violates FOCUS PROMINENCE, since the
phonological phrase that contains the focus is not the head of the intonational
phrase, and the latter violates iP-Hd-R, since the phonological phrase that is the
head of the iP is not right-aligned with it. But with the ranking PrWd=pP >> SO,
candidate (37a), which sacrifices optimal phrasing, now loses to the candidate
that sacrifices
canonical
constituent
order
(more
concretely,
structural
prominence of the subject over the object) instead.
Lastly, consider German, which as proposed allows for either deviating form
the optimal phrasing pattern or from canonical constituent order.
(38) a. Who’s in charge of nominating the foreign minister?- It is expected
that…
b. der KANzler den Aussenminister ernennt.
SOV
the chancellor the foreign-minister nominates
‘…the chancellor nominates the foreign minister.’
c. den Aussenminister der KANzler
ernennt.
OSV
the foreign-minister the chancellor nominates
‘…that the chancellor nominates the foreign minister.’
In this case the correct result is arrived by means of a constraint tie between
SO and XP=pP, the constraint that dictates optimal phrasing in this language.
The resulting analysis is presented in tableau (39), where the constraint tie is
represented by not having any division between the tied constraints in the
topmost row of the tableau.
25
(39) German subject focus
FOCP
F a. (der KANzlerFoc den Aussenminister
ernennt).
b. (der KanzlerFoc) (den AUssenminister
ernennt).
c. (der KANzlerFoc) (den Aussenminister
ernennt).
F d. (den Aussenminister) (der KANzlerFoc
ernennt).
iPHd-R
XP=pP
SO
*
*!
*!
*
Consider now cases of sentence focus. As mentioned in the introduction, this
is an instance where all three languages show a similar behavior; not only does
the nuclear accent fall on the direct object, and the felicitous answers show
canonical constituent order, but also the winning candidates in the tableaux (35),
(37), and (39) (i.e. the optimal outputs for cases of subject focus) are all
infelicitous answers to the relevant wh-question.
(40) English
a. What’s been happening?
b. John bought the NEWSpaper.
c. #JOHN bought the newspaper.
(41) Spanish
a. Qué pasó?
‘What happened.’
b. Juan compró ayer
el peRIÓdico.
Juan bought yesterday the newspaper
‘Juan bought the newspaper yesterday.’
c. #Ayer
compró el periódico JUAN.
yesterday bought the newspaper Juan
(42) German
a. Worauf
warten alle? -Es wird erwartet, dass
where-upon wait
all it is
expected that
What’s everybody waiting for? -It is expected that…
26
b. der Kanzler
den AUssenminister ernennt.
the chancellor the foreign-minister nominates
‘… the chancellor nominates the foreign minister.’
c. #der KANzler den Aussenminister ernennt.
the chancellor the foreign-minister nominates
‘…the chancellor nominates the foreign minister.’
d. #den Aussenminister der KANzler ernennt.
the foreign-minister the chancellor nominates
‘…that the chancellor nominates the foreign minister.’
The data in (40-42) can be captured straightforwardly in the analysis
developed so far. We assume that the subject, the verb and the object are all
F-marked, rather than just the highest constituent, say S. In this we follow Selkirk
(1984, 1995), among many others, where it is convincingly argued that such
subordinated F-markers are crucial in accounting for deaccenting phenomena
within the focus, though we must omit discussion of these cases in the present
paper in the interest of space (cf. Büring 2001a). Consequently, FOCUS
PROMINENCE will necessarily be violated, since only one of the three can get to be
the head of the phonological phrase that will in turn be the head of the
intonational phrase.17 Since there is no way to avoid these violations of FOCUS
PROMINENCE anyway, candidates that depart from canonical phrasing and
constituent order (40c, 41c, 42c, d) will be less harmonious than those that do not
violate SO and the highest ranked phrasing constraint that dictates optimal
phrasing. This analysis is presented in the tableaux in (43-45).
(43) English sentence focus
FOCP
a. (JOHNFoc boughtFoc the newspaperFoc)
F b. (JohnFoc)(boughtFoc the NEWSpaperFoc)
c. (JOHNFoc)(boughtFoc the newspaperFoc)
d. (boughtFoc the newspaperFoc)(JOHNFoc)
**
**
**
**
iP-HdR
SO
XP=pP
*!
*!
*!
27
(44) Spanish sentence focus
FOCP
a. (JUANFoc compróFoc el periódicoFoc)
F b. (JuanFoc)(compróFoc)(el peRIÓdicoFoc)
c. (JUANFoc)(compróFoc)(el periódicoFoc)
d. (compróFoc)(el periódicoFoc)(JUANFoc)
iP-HdR
**
**
**
**
PrWd=pP
SO
N!V
*!
*!
(45) German sentence focus
FOCP
a. (der KANzlerFoc den AussenministerFoc
ernenntFoc).
F b. (der KanzlerFoc) (den AUssenministerFoc
ernenntFoc).
c. (der KANzlerFoc) (den AussenministerFoc
ernenntFoc).
d. (den AussenministerFoc)(der KANzlerFoc
ernenntFoc).
iPHd-R
**
XP=pP
SO
*!
**
**
**
*!
*!
The crucial observation is that, in contrast to what is observed in cases of
constituent focus, sacrificing canonical phrasing or canonical constituent order
does not improve the structures in any respect (i.e. the violations FOCUS
PROMINENCE of cannot be avoided either way), so the canonical structures first
presented in (25) emerge as the winners. As the reader will be able to verify, the
rankings in (33) also account for those cases in (4-6) where the VP or the direct
object is the focus of the sentence.
4. Conclusions
In this paper, we have provided a constraint-based cross-linguistic analysis
of focus-related constituent order in English, Spanish and German based on the
theories of prosodic phrasing of Selkirk 1995, Truckenbrodt 1999, inter alia. We
have shown how this kind of analysis can preserve the insights captured in
Zubizarreta's (1998) work on constituent order, without resorting to a mechanism
28
by which intonational prominence is derived syntactically (the NSR), and without
assuming different intonational and syntactic properties for the languages under
consideration.
The core of this analysis has been to claim that languages differ in whether
they sacrifice phonological phrasing requirements or syntactic requirements
when there is a potential conflict between FOCUS PROMINENCE and iP-Hd-R, or
whether they allow for both possibilities. By making use of ranked violable
constraints, the analysis further captures the parallelism that all three languages
display when there is no conflict between these two constraints or when violation
of FOCUS PROMINENCE is altogether unavoidable.
R EFERENCES
Ackema, Peter (2001). On the relation between V-to-I and the structure of the
inflectional paradigm. The Linguistic Review 18, 233-263.
Alcoba, Santiago and Murillo, Julio. (1998). Intonation in Spanish. In Intonation
systems: a survey of 20 languages, Daniel Hirst and Albert Di Cristo (eds),
152-166. Cambridge, Cambridge University Press.
Bolinger, Dwight. (1954). English prosodic stress and Spanish sentence order.
Hispania 37, 152-156.
Bolinger, Dwight (1965). Forms of English: Accent Morpheme, Order.
Cambridge, MA: Harvard University Press.
Büring, Daniel (2001a). Let's Phrase It! - Focus, Word Order, and Prosodic
Phrasing in German Double Object Constructions. In Competition in Syntax,
Müller, Gereon and Wolfgang Sternefeld (eds.), 101-137. Berlin & New
York: de Gruyter.
Büring, Daniel (2001b). What do Definites do that Indefinites Definitely don't?
In Audiatur Vox Sapientiae: AFestschrift for Arnim von Stechow, Fery,
29
Caroline and Wolfgang Sternefeld (eds.) studia grammatica 52, 70-100.
Berlin: AkademieVerlag.
Choi, Hye Won. (1999). Optimizing Structure in Context: Scrambling and
Information Structure. Stanford, CSLI.
Contreras, Heles. (1976). A theory of constituent order with special reference to
Spanish. Amsterdam: North Holland.
Contreras, Heles. 1991. On the position of subjects. In Perspectives on Phrase
Structure, Susan Rothstein (ed.), Syntax and Semantics 25, 69-73. San
Diego, Academic Press.
Costa, João. (1996).
Constituent order
and constraint interaction. Ms.,
HIL\Leiden University; to appear in Seminários de Linguística, University of
Algarve.
Costa, João. (1998). Word Order Variation: A Constraint Based Approach.
Unpublished Doctoral Dissertation, Leiden University.
Face, Timothy. (2000). A phonological analysis of rising pitch accents in
Castilian Spanish. Paper presented at the 30th Linguistic Symposium on
Romance Languages, University of Florida, Gainesville.
Fant, Lars. (1984). Estructura informativa en español: estudio sintáctico y
entonativo. Stockholm: Almqvist & Wiskell.
Frascarelli, Mara (2000). The Syntax-Phonology Interface in Focus and Topic
Constructions in Italian. Studies in Natural Language and Linguistic Theory
50. Dordrecht/ Boston/ London: Kluwer.
Grimshaw, Jane (1997). Projection, Heads, and Optimality. Linguistic Inquiry
28(3), 373-422.
Gussenhoven, Carlos (1983). Focus, Mode, and the Nucleus. Journal of
Linguistics 19, 377-417. Reprinted as chapter 1 in Gussenhoven (1984).
Gussenhoven, Carlos (1984). On the Grammar and Semantics of Sentence
Accents. Dordrecht: Foris.
30
Gutiérrez-Bravo, Rodrigo. (2000). Focus, phonological phrasing and word order
in Spanish and English. Ms. University of California, Santa Cruz.
Gutiérrez-Bravo, Rodrigo.
(2002).
Structural
Markedness
and
Syntactic
Structure: A Study of Word Order and the Left Periphery in Mexican
Spanish. Unpublished doctoral dissertation, University of California, Santa
Cruz.
Hayes, Bruce and Lahiri, Aditi. (1991). Bengali intonational phonology. Natural
Language and Linguistic Theory 9, 47-96.
Hualde, José Ignacio. (2002). Intonation in Spanish and other Ibero-Romance
languages: Overview and status quaestionis. In Romance Phonology and
Variation: Selected Papers form LSRL XXX, Wiltshire, Caroline and Camps,
Joaquim (eds.), 101-115. Amsterdam/Philadelphia, John Benjamins.
Jackendoff, Ray (1972). Semantics in Generative Grammar. Cambridge, MA:
MIT Press.
Jacobs, Joachim (1993) Integration. In Wortstellung und Informationsstruktur,
Reis, Marga (ed.), 64-116. Tübingen: Niemeyer.
Ladd, D. Robert. (1980). The Structure of Intonational Meaning. Bloomington:
Indiana University Press.
Ladd, D. Robert. (1996). Intonational Phonology. Cambridge, Cambridge
University Press.
Lenerz, Jürgen. (1977). Zur Abfolge Nominaler Satzglieder im Deutschen.
Tübingen: Narr
McCarthy, John. (2002). A Thematic Guide to Optimality Theory. Cambridge,
Cambridge University Press.
Müller, Gereon. (1998). German Constituent order and Optimality Theory.
Arbeitspapier des SFB 340, #126.
31
Neeleman, Ad and Reinhart, Tanya. (1998). Scrambling and the PF interface. In
The Projection of Arguments, Geuder, Wilhelm and Butt, Miriam (eds.), 309353. Stanford, CSLI.
Ordóñez, Francisco and Treviño, Esthela. (1999). Left-dislocated subjects and
the pro-drop parameter: A case study of Spanish. Lingua 107. 39-68.
Rizzi, Luigi. (1997). The fine structure of the Left Periphery. In Elements of
Grammar, L. Haegeman (ed.), 281-337. Dordrecht, Kluwer.
Rochemont, Michael (1986). Focus in Generative Grammar. Amsterdam/
Philadelphia: John Benjamins.
Samek-Lodovici, Vieri (2002). Prosody-Syntax Interaction in the Expression of
Focus. Ms. University College London.
Schmerling, Susan F. (1976). Aspects of English Sentence Stress. Unpublished
doctoral dissertation, University of Texas at Austin.
Schwarzschild, Roger. (1999). GIVENness, AvoidF and other constraints on the
placement of accent. Natural Language Semantics 7: 141-177.
Selkirk, Elisabeth. (1984). Phonology and Syntax: the relation between sound
and structure. Cambridge, Mass.: MIT Press.
Selkirk, Elisabeth. (1995). Sentence prosody: Intonation, Stress and Phrasing. In
The Handbook of Phonological Theory, Goldsmith, J. (ed.), 550-569.
Oxford: Blackwell.
Silva-Corvalán, Carmen. (1983). On the interaction of Constituent order and
Intonation: some OV Constructions in Spanish. In Discourse Perspectives on
Syntax, Klein-Andreu, Flora (ed.), 117-140. New York: Academic Press.
Sosa, Juan Manuel. (1991). Fonética y fonología de la entonación del español
hispanoamericano. Unpublished
doctoral
dissertation, Department of
Spanish and Portuguese, University of Massachusetts, Amherst.
Sosa, Juan Manuel. (1999). La entonación del español: Su estructura fónica,
variabilidad, y dialectología. Madrid, Cátedra.
32
Steedman, Mark (1991). Structure and Intonation. Language 67, 260-296.
Szendröi, Kriszta (2001).
Focus
and
the
Syntax-Phonology
Interface.
Unpublished doctoral dissertation, University College London.
Truckenbrodt, Hubert. (1995). Phonological Phrases: Their Relation to Syntax,
Focus and Prominence. Unpublished doctoral dissertation, MIT.
Truckenbrodt, Hubert. (1999). On the relation between syntactic phrases and
phonological phrases. Linguistic Inquiry 30, 219-255.
Vallduví, Enric. (1992). The informational component. New York, Garland Press.
Vikner, Sten .(1995). Verb Movement and Expletive Subjects in the Germanic
Languages. Oxford, Oxford University Press.
Webelhuth, Gert .(1992). Principles and Parameters of Syntactic Saturation.
Oxford, Oxford University Press.
Zubizarreta, María Luisa. (1998). Prosody, Focus and Constituent order.
Cambridge, Mass.: MIT Press.
NOTES
*
We would like to thank Judith Aissen, Heriberto Avelino, Junko Itô, Armin
Mester, Anne Sturgeon, the participants at Joan Bresnan’s and Judith Aissen’s
OT Syntax seminars (Stanford University, Winter 1999 and UCSC, Winter 2000,
respectively) and the audience at the Topic and Focus Workshop at Going
Romance 2000 for helpful comments and discussion of the material presented
here. The authors are responsible for any remaining errors. R. Gutiérrez-Bravo’s
research was supported in part by the NSF grant SBR-9818177, and by the
National Council of Science and Technology
of Mexico (CONACYT),
scholarship No. 177325. An earlier version of this paper appeared in Syntax and
Semantics at Santa Cruz 3, James McCloskey (ed.), University of California at
Santa Cruz, Linguistics Research Center, 41-58.
33
1
Throughout this paper we use German subordinate clauses to keep the
constituent order effects independent from the verb-second
requirement
observed in matrix clauses. Since constituent order in German is influenced not
only by focus considerations, but also by the relative animacy and definiteness of
the arguments of the verb involved (Lenerz 1977, Müller 1998), both the subject
and the object display the same degree of animacy and definiteness in the
German examples in order to keep the focus phenomena independent from these
other variables.
2
At the time of the revision of this article we can add to this Szendröi (2001)
and Samek-Lodovici (2002), which are very similar spirit, as well as Frascarelli
(2000), which presents a prosody based account that does not use ranked
violable constraints. Since we didn’t have access to these works before, we will
not comment on them in the rest of this article.
3
Following Vikner (1995) we also assume that V moves to I in German. This,
however, has no effects on the relative constituent order of the subject and the
object since the verb moves to the head position of a right-headed IP (see also
Webelhuth 1992). Finally, we assume that V-to-I movement is governed by an
independent
constraint
related
to
the
morphology
of
the
verb,
the
characterization of which will not be relevant for our analysis (see Grimshaw
1997 and Ackema 2001 for some proposals).
4
Notice, in the absence of a direct object, movement of the subject to Spec-IP is
not forced by SO (thanks to João Costa for bringing this point to our attention).
On a first impression this would seem to be an undesirable consequence, but
upon closer inspection, the situation is not so clear. English, for example, despite
its fairly rigid constituent order, still has a good number of constructions where
the subject of an intransitive verb does not move to Spec-IP. Such is the case of
locative inversion (In that corner now stands a mini-mall) and expletive-associate
34
constructions (There are many students in that class). Crucially, these kinds of
constructions are not attested in English when the verb is transitive. Although this
issue certainly merits further investigation, we assume that the structural
condition stated in the SO constraint is a relevant requirement, even when there
may be more than one constraint at play in deriving movement of the subject to
Spec-IP.
5
As the reader will be able to verify later, this assumption is not necessary to
derive the word order facts observed in transitive sentences, which could also be
derived by an analysis where Spanish has the same pattern of phonological
phrasing as English and German (many thanks to Judith Aissen and Anne
Sturgeon for bringing this point to our attention). Constituent order variation in
intransitive clauses, however, provides evidence that the pattern of phonological
phrasing is the one we suggest here. Concretely, the requirement that the focus
be sentence final is also observed in clauses with intransitive verbs in Spanish, as
is well know. This is exemplified by the following contrast between Spanish and
English, adapted from Bolinger (1954).
(i) a. The fighting will be over as soon as a TRUCE is signed.
b. They’ve been talking truce for two years, but I wont believe anything
until the truce is SIGNED.
(ii) a. El conflicto terminará tan pronto ![se firme una TREGUA].
the conflict willl.finish as soon be-signed a
truce.
‘The conflict will be over as soon as a TRUCE is signed.’
VS
b. Desde hace dos años vienen
hablando de una tregua, pero yo
from madetwo years they.come talking of a truce but I
no voya creer
nada
hasta que [la tregua se FIRME]. SV
not going to believe nothing until that the truce be-signed
‘For two years they have been talking about a truce, but I won’t
believe anything until the truce is SIGNED.!’
If Spanish had the same pattern of phonological phrasing as English, we
would expect it to resort to the strategy of accenting the focus in-situ
35
characteristic English (i.a-b), and not to the word order variation observed in (ii),
contrary to fact. In contrast, in our analysis the word order variation in (ii) is
expected. This is because each prosodic word forms its own phonological phrase
in this language. Consequently (as will be discussed in detail in what follows), the
prosodic word that is the focus must appear in the sentence final position if the
phonological phrase that corresponds to this prosodic word is to be the one that
receives the nuclear accent.
6
It is worth noting that in contradistinction to these theories, and in line with
Truckenbrodt (1999), this accent domain formation is not viewed by us as a
process pertaining to the realization of focus in particular, but to prosodic
phrasing in general. Accordingly, we expect it to occur with predicate-argument
sequences in the background as well, which is correct. Thus in the answer in (i),
the same distribution of accents is found within the relative clause, which is part
of the background, as in the (focused) main clause:
(i) (What did the girl that wore the hat buy?) The girl that wore the HAT
bought a NEWSpaper.
7
See also Schwarzschild (1999) for a similar constraint. Once again, the effect of
A/P follows from Truckenbrodt’s (1999) STRESS-XP constraint, but A/P further
captures all instances of predicate-argument integration shown in (11), which
STRESS-XP does not.
8
Fant (1984) uses the A+ notation to indicate that the leftmost A accent is the
most prominent A accent.
9
The gloss and the free translation are our own.
10
For concreteness, we assume that iPhd-R outranks A/P in both English and
German. In the case of German, for instance, justification for this ranking can be
36
found in cases of narrow verb focus, which, however, we cannot discuss because
of space considerations. The ranking iPhd-R >> XPpP of English and German
will be argued for in detail in section 4 of this paper.
11
The reader will be able to note that the ranking iPHd-R >> PrWd=pP in
Spanish is not crucial for the analysis of the examples that follow. We have
assumed this ranking, however, because the data presented in most works on
Spanish intonation cited here point to the conclusion that iPHd-R is undominated
in Spanish, whereas PrWd=pP is not. In example (19), for instance, two prosodic
words, the verb and the direct object, presumably form a single phonological
phrase, in violation of PrWd=pP.
12
But see Face (2000) for an alternative interpretation of these data.
13
Rizzi (1997) reports this same property for fronted foci in Italian.
14
This version of the constraint, from Büring (2001a), takes both the focus and
the domain of the focus to be prosodic units, and defines the notion of being a
focus as containing an F-marked syntactic node. This is in all likelihood the
intention behind Truckenbrodt’s (p.107) formulation “If F is a focus and DF is
its domain, then the highest prominence in DF will be within F.”
15
An anonymous reviewer asks whether it is not a problem for our analysis that
losing candidate (35d) corresponds to an ungrammatical sentence, while this is
not the case of the other losing candidates in this tableau: candidate (35b), for
instance, is just infelicitous in this particular context. This does not represent a
problem in the characterization of the distinction between ungrammaticality and
infelicity in OT in Gutiérrez-Bravo (2002), which is the one we adopt here (see
also McCarthy 2002). In this characterization, ungrammatical and infelicitous
candidates are not inherently different. An ungrammatical candidate is a
structure that is not optimal for any input, given the grammar (i.e. the constraint
ranking) of the language. In contrast, a candidate that is perceived as infelicitous
37
in a specific context is a structure that is not optimal for the input under
consideration, but which (again, given the constraint ranking of the language) is
optimal for at least one input. The losing candidate (35b), for instance, actually
corresponds to the felicitous utterance in cases of sentence focus. As such, the
difference between ungrammatical and infelicitous candidates does not have to
do with their inherent properties (and so it is valid to compare them in a tableau),
but rather with the relation they have with the set of inputs and the constraint
ranking of a given language.
16
In the tableau in (37), we leave out for clarity the temporal adverb ayer
‘yesterday’, in the different structures in the candidate set. This has no effect on
the evaluation of the candidates (even when the absence of ayer in the preverbal
position in (36c) leads to ungrammaticality in some varieties of Spanish, such as
Mexican Spanish) since the presence of the adverb plays no role in the
satisfaction or violation of the constraints in our analysis.
17
Alternatively, each prosodic word could form its own pP – (JOHN)(BOUGHT)(the
NEWSPAPER)
- making each focused word maximally prominent within its pP. But
this would yield the exact same violation profile at the iP level because only one
of these pPs could in turn become maximally prominent within the iP (due to the
same logic, (43b) and (45b) have as many FocP violations as (43a) and (45a),
respectively; one because the verbal PWd is not the head of its pP, the second
because the pP corresponding to the subject is not the head of iP). Ultimately, all
but one focus will be prosodically subordinated, in violation of FocP; whether
this happens at the pP or at the iP level depends on whether any other constraint
like XP=pP favors larger or smaller pPs.
Note, too, that FocP violations are counted per prosodic unit. Thus, a focus
marker on the VP will not cause additional violations in, say, (43c) since the
smallest prosodic unit containing it is the second pP, i.e. the same that contains
38
the verbs and the object. The same applies mutatis mutandis to F-marks on the S,
except that in all the cases discussed here, the smallest prosodic unit containing S,
iP, is maximally prominent. Our treatment is thus fully compatible with Selkirk’s
theory of F-marking, or its recent adaptation in Schwarzschild (1999).